Episode 3 · Goodhart's Law
Busier, not richer.
You sell candles online. You set a goal of 100 orders a month, cut the price, offered free shipping, and hit it. Twice the orders, twice the hours, and you kept $250 instead of $650.
Free cheat sheet
Goodhart's Law on one page
How a target drifts, how to pair a number with one it can't fool, every number from the episode, when it breaks, and one question to ask this week. Print it or keep it open in a tab. No sign-up.
The idea
When a number becomes your goal, you'll find ways to hit it that miss the point. A number is a good measure while nobody is pushing on it. Make it a target, and you (like everyone) reach for the cheapest way to move it. The economist Charles Goodhart noticed this in 1975, writing about monetary policy in the U.K.
The anthropologist Marilyn Strathern put it in one line in 1997:
"When a measure becomes a target, it ceases to be a good measure."
In the episode, orders used to tell you how many people wanted your candles. Once 100 orders became the goal, the fastest ways to get there were 25% off and free shipping. The number went up. The money went down. You didn't cheat. You did exactly what the number asked.
The fix: pair it with a number it can't fool
- Name the number you watch most. Orders, a budget line, a streak, a score.
- Ask the question. What's the easiest way to hit this number without getting what you actually want? For orders, it's selling at a price where you barely keep anything.
- Add a second number that would catch it. Goodhart's Law is why good measures come in pairs. Orders, checked by dollars kept per order. A grocery budget, checked by what you spent on every card. A no-spend streak, checked by your balance a month later. A credit score, checked by the total you owe.
- Watch the pair, not the one. If the first number moves and the second doesn't, you're gaming yourself.
Worked example from the episode
| Step | Math | Answer |
|---|---|---|
| Before the goal: $24 a candle, the buyer pays shipping. $9 to make it, about $2 in fees | $24 − $9 − $2 | $13 kept per order |
| 50 orders a month, 10 hours of packing | 50 × $13 | $650, $65 an hour |
| Goal month: 25% off ($18) and free shipping, which costs you $5. Fees drop to $1.50 | $18 − $9 − $5 − $1.50 | $2.50 kept per order |
| 100 orders, goal hit, 20 hours of packing | 100 × $2.50 | $250, $12.50 an hour |
| Next month: back to $24, the buyer pays shipping, watching both numbers | 70 × $13 | $910 in 14 hours, $65 an hour |
The candle shop is a story. Examples are illustrative numbers, not real data, and nothing here is financial advice. Next month misses the goal of 100 and makes more money than ever.
When it breaks
- Numbers aren't the enemy. It's a tendency, not a law of physics. Without counting orders, you'd never have caught this.
- Watching is not chasing. The danger is one number with a lot riding on it. A number you watch to learn from keeps its meaning.
- Volume can be the right call. If you choose it on purpose. A launch discount to earn your first reviews can make sense, with a set cost and an end date.
- No pair is game-proof. Every pair of numbers leaves some way to game it. When you see the gaming, change the measure and look at the real work behind it.
Try this week
Open one app you check often: a budget, a bank, a fitness tracker, a shop dashboard. Write down the one number you watch most. Then ask yourself: what's the easiest way to hit this number without getting what you actually want? Write down a second number that would catch it.
And the next time someone shows you a great number, ask:
"What did it cost to hit it?"
Sources: C. A. E. Goodhart, "Problems of Monetary Management: The U.K. Experience," Reserve Bank of Australia, 1975 (printed 1976). Marilyn Strathern, "'Improving ratings': audit in the British University system," European Review 5(3), 1997. New to the channel? Start with Episode 1, Little's Law, then Episode 2, Theory of Constraints.